"Jan-Peter Seifert" <Jan-Peter(dot)Seifert(at)gmx(dot)de> wrote:
> I wonder whether extra measures are necessary to recover from a
> disk full error besides freeing enough disk space?
> Is it no problem if the WAL is within the same disk space and
> can't be written as well?
> Should you free enough disk space, do a checkpoint, restart the
> server and check the server log?
At our shop, we have rarely run out of disk space on a production
machine, due to close monitoring; but in those few cases, and in the
somewhat more frequent case that it happened in a development
environment, we have simply freed space and continued. From what I
have seen, the PostgreSQL development community tries very hard to
ensure that you can do this safely. On the other hand, it would
obviously be very hard to thoroughly test every possible code path
under such conditions, so I always try to watch for possible
problems afterward.
-Kevin